The challenge
Cost per acquisition was rising. The airline's own customer data was not being used well.
What we did
We trained a model on the airline's own customer and booking data. It scored each person on how likely they were to book. Bids went up for likely travellers and down for the rest. We changed the goal from a lower CPA to more booked journeys per dollar. Ads then moved people from inspiration to price to flight offers.
The result
Bookings doubled against a 40% target. Cost per booking (CPA) fell about 20%, double the goal. Use of first-party data rose 45%. The bidding model now keeps growth going.
